Molecular and clinical features of basal-like, claudin-low, and triple-negative breast cancers^ Molecular profiling studies of breast cancer have firmly established the existence of biologically distinct subtypes of breast cancer, including basal-like, luminal, HER2, and normal breast like'^. HER2 or basal-like tumors comprise about 10% and 15% of all breast cancers, respectively, and are characterized by poor prognosis^''^. This is in part due to the propensity of these subtypes to metastasize. For basal-type breast cancer, this might result from the resemblance of these tumor cells to myoepithelial cells of the breast and the expression of factors promoting proliferation, migration and invasion, as well as suppression of apoptosis ^. Recently, a claudin-low breast cancer subtype was identified which shares many features with the Basal B subtype identified by us before^'^. Claudin-low tumors are characterized by loss of cell-cell adhesion molecules, again promoting a metastatic phenotype, and are enriched for stem-cell like features, which might contribute to resistance to chemotherapy^'^. Basal-like and claudin-low breast cancers demonstrate loss of expression hormone receptors and HER2, markers that are used clinically to subtype breast cancers. These triple-negative breast cancers (TNBC) lack expression of targets for targeted therapies and are therefore only amenable to chemotherapy, which has only limited activity and the median survival for patients with TNBC remains low at 13.3 months following diagnosis of metastatic disease^. Very recently, PARP inhibitors have shown significant activity in TNBC patients, the first targeted therapy for this subtype^^. Since claudin-low tumors represent a recently described subgroup of TNBCs and only limited information on their biology is available, we will focus the experimental part of this proposal on this clinically challenging group of breast cancers. The MAPK pathway in TNBC and the role of feedback loops^ At the molecular level, basal-like and claudin-low breast cancers are characterized by strong expression of genes comprising a proliferation signature and expression of growth factor receptors, including EGFR and cKit. In addition, molecules mediating signal transduction from these receptors, including MEK, ERK, and PIS kinase (PISK) have been demonstrated to be upregulated in this subtype^^'^. At the same time, the Rb pathway is frequently mutated as is p5S, thus further enhancing the proliferative capacity of this tumor type and preventing apoptosis. By taking a systems biological approach (funded by the NCI CCSB), we found that indeed basal-like breast cancers appear to preferentially utilize the MAPK pathway and, in agreement with this, are particularity sensitive to small-molecule inhibitors of MEK in terms of inhibition of cell proliferation. We also discovered that a MEK-EGFR-PISK feedback loop contributes to resistance to MEK inhibitors (see Progress Report). Thus, cellular signal transduction pathways are complex, highly interconnected and dynamic networks which are characterized by cross-talk and feedback mechanisms^. Among the critical factors mediating such negative feedback loops are phosphatases such the dual-specificity phosphatases that are induced following EGFR simulation. The clinical relevance of negative feedback loops is highlighted by the cell-protective activation of AKT that occurs following mTOR inhibition (mediated by insulin-receptor substrate-1 (IRS-1)) in tumors with constitutively active mTOR signaling^. Similarly, activation of MAPK through a feedback loop meditated by S6 kinase, PIS kinase, and Ras following mTOR inhibition has been found and addition of MEK inhibitors enhanced the anti-tumor effect of mTOR inhibitors in vitro and in vivo^. In addition, swapping of receptor kinases following inhibition of one receptor in cancer cells with multiple activated RTKs has been demonstrated. Thus, feedback mechanisms are capable of maintaining signaling in redundant systems. In the case ofthe MEKEGFR_ PI3K feedback loop described above, it remains unclear which molecular mechanisms mediate this effect. Because of the important role of feedback loops, we will investigate those mechanisms as part of this proposal.